Journal of Physics: Conference Series

PAPER • OPEN ACCESS Related content

- kspectrum: an open-source code for high-
Delft Research Controller: an open-source and resolution molecular absorption spectra
production
community-driven wind turbine baseline controller V. Eymet, C. Coustet and B. Piaud

- Turbulence Impact on Wind Turbines:

Experimental Investigations on a Wind
To cite this article: SP Mulders and JW van Wingerden 2018 J. Phys.: Conf. Ser. 1037 032009 Turbine Model
A Al-Abadi, Y J Kim, Ö Ertunç et al.

- Simulation modelling for new gas turbine

fuel controller creation.
L E Vendland, V G Pribylov, Yu A Borisov
View the article online for updates and enhancements. et al.

This content was downloaded from IP address 185.22.154.160 on 15/12/2018 at 15:48

The Science of Making Torque from Wind (TORQUE 2018) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1037 (2018)
1234567890 ‘’“” 032009 doi:10.1088/1742-6596/1037/3/032009

Delft Research Controller: an open-source and

community-driven wind turbine baseline controller
SP Mulders and JW van Wingerden
Delft Center for Systems and Control, Faculty of Mechanical Engineering, Delft University of
Technology, Mekelweg 2, 2628 CD Delft, The Netherlands
E-mail: S.P.Mulders@tudelft.nl, J.W.vanWingerden@tudelft.nl

Abstract. Wind energy research groups from various disciplines generally use self-developed
baseline wind turbine control implementations and tunings, which complicates the evaluation
and comparison of new control algorithms. To solve this problem, the Delft Research Controller
(DRC) provides an open, modular and fully adaptable baseline wind turbine controller to the
scientific community. New control implementations can be added to the existing baseline
controller, and in this way, convenient assessments of the proposed algorithms is possible.
Because of the open character and modular set-up, scientists are able to collaborate and
contribute in making continuous improvements to the code. The DRC is being developed
in Fortran and uses the Bladed-style DISCON controller interface. The compiled controller is
configured by a single control settings parameter file, and can work with any wind turbine model
and simulation software using the DISCON interface. Baseline parameter files are supplied for
the NREL 5-MW and DTU 10-MW reference wind turbines.

1. Introduction
The existence of reference models and baseline cases is a crucial aspect in the scientific
community, as it allows for convenient and fair evaluation of proposed innovations. In the
wind turbine community, the National Renewable Energy Laboratory (NREL) offshore 5-MW
baseline wind turbine is a fictive but fully defined reference wind turbine (RWT) model [1], and
is actively used in the scientific field. To accommodate the next step in enlarging the size and
rated power of offshore wind turbines, the Technical University of Denmark (DTU) provides a
10-MW reference wind turbine model [2]. The DTU10MW model is developed in cooperation
with Vestas Wind Systems. Its design is mainly focused on setting a reference for next generation
rotors, with good aerodynamic performance at a relative low weight.
Besides reference models, wind turbine simulation software is also largely standardized. The
industry standard, commercial and certified high-fidelity wind turbine simulation package is
Bladed by DNV GL [3]. On the other hand, an open-source aeroelastic package for simulating
horizontal-axis wind turbines is FAST, which was up until recently actively developed and
maintained by NREL. However, a shift towards community-driven software development is seen
in the scientific field, and OpenFAST is established with the FAST v8 code as its starting point
[4]. The goal of OpenFAST is being a community model, with users and developers from research
laboratories, academia and industry improving software quality and accelerating development.
In contrast to the previously mentioned reference models and simulation software, no clear
choice of a baseline wind turbine controller currently exists that is easy to use, modular and

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
The Science of Making Torque from Wind (TORQUE 2018) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1037 (2018)
1234567890 ‘’“” 032009 doi:10.1088/1742-6596/1037/3/032009

extendable. Wind energy research groups from various disciplines generally use self-developed
baseline control implementations and tunings, of which the source code is rarely available. This
negatively impacts the ability to compare results from different research projects or groups. It
has to be noted that NREL provides an open-source controller for its NREL5MW reference
wind turbine [1]. However, this controller is limited in functionality and inconvenient to extend
or interchange between distinct wind turbine models as functionality, turbine parameters and
controller tunings are hard-coded in a single source file. DTU also provides an internally
developed controller for their DTU 10-MW RWT [5] of which the source code is available,
but the development is not driven by a broad community.
To this end, the Data Driven Control wind energy research group from Delft University
of Technology started the initiative to develop an open-source and community-driven wind
turbine baseline controller. A design specification was that the controller should be generally
applicable to all turbine models defined in simulation software that uses the Bladed-style
DISCON controller interface [6], such as OpenFAST, Bladed or HAWC2. Also, a convenient
way of configuring the controller should be present, without editing the source code and thus
the need for recompilation. With these goals in mind, the foundations of a baseline wind turbine
controller have recently been laid out, and is dubbed the Delft Research Controller (DRC). For
consistency with OpenFAST, the DRC is being developed in the Fortran programming language
(free-form) [7]. The modular and open character allows scientists to collaborate and contribute
in making continuous improvements to the code. The DRC is provided with a toolbox consisting
of regularly used (control) functions and filters to allow rapid development and implementation
of new contributions. Remarkable functionality of the DRC is the ability to activate and switch
between conventional yaw-rate [8] and novel yaw-by-IPC control implementations [9].
The main contribution of this paper is to provide a comprehensive description of the DRC.
The organization of this paper is as follows. In Section 2, an overview of the DRC is given
including the built-in function and filter modules. These components are used to make up the
baseline torque, (individual) pitch and yaw controllers, described in Section 3. In Section 4,
simulations of the DTU10MW controlled by the DRC showcase the performance of various
control options. In Section 4, conclusions are drawn and an outlook is given on future additions.

2. DRC components
This section gives an overview and description of the various components included in the DRC.
A general overview of the controller is given in Section 2.1, after which the included filters
and functions are described in Sections 2.2 and 2.3, respectively. In the remainder of this paper,
internal controller variables, data-types, modules and subroutines are presented in a typewriter
font, whereas parameters defined in the controller configuration (parameter)file are underlined.

2.1. Overview of the DRC

This section gives an overview of the DRC, of which the overall working principle is presented in
Figure 1. The DRC is set up such that only a single parameter file ControllerParameters.in
for each wind turbine model is required to define the complete control system. This feature
removes the need for repetitive recompilation of the controller under a change in control
settings. To prevent computational errors related to unit conversions, all variables and constants
are defined in SI units. The input file allows defining different debugging levels: setting
LoggingLevel to 0 omits writing a log file, defining it as 1 writes specified variables to a
.dbg file, and configuring the parameter to 2 additionally writes the complete avrSWAP-array to
a .dbg2 file.
The DRC consists of multiple modules containing commonly used functions and subroutines.
The Constants.f90 module contains regularly used non-varying parameters, such as unit

2
The Science of Making Torque from Wind (TORQUE 2018) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1037 (2018)
1234567890 ‘’“” 032009 doi:10.1088/1742-6596/1037/3/032009

Delft Research Controller (DRC) OpenFAST, Bladed,

HAWC2, ...

Controller Functions and Derived data types ReadAvrSwap()

Parameters.in Subroutines

SetParameters()

avrSWAP(*)
ControlPar StateMachine()
Wind turbine
Constants.f90

Functions.f90

LocalVar model
Filters.f90

<Simulation
ObjectInst VariableSpeedControl()
Name>.dbg

PitchControl()

OptionalControl()

Figure 1. A schematic of the DRC architecture for wind turbine control. The controller
exchanges data using the so-called the DISCON external controller interface via the avrSWAP-
array. The DRC is completely parameterized by a single configuration file, and writes debug
information to a log file when desired.

conversions and physical constants. The Functions.f90 module contains general procedures,
and the Filters.f90 module consists out of discretized signal filters.
The DRC uses derived data types to store parameters and variables in a centralized manner.
A derived data type has similarities to C and MATLAB structures, or C++ classes. The values
defined in the ControllerParamameters.in file are loaded in the ControlPar data type and
are stored as read-only. All local controller variables are stored in the LocalVar data type, such
that they are accessible among the different control subroutines.
The DRC executes function calls in a fixed sequence during each control iteration. The DRC
reads the control in- and output avrSWAP-array by calling its ReadAvrSwap() subroutine, and
the SetParameters() subroutine evaluates required controller parameters and performs value
assertions. Next, before calling any controller, the state-machine subroutine StateMachine()
determines the state of the turbine, and this information is used by the controllers to perform
corresponding control actions. To enable the Variable-Speed Variable-Pitch (VSVP) control
strategy, torque and pitch control subroutines (VariableSpeedControl() and PitchControl())
are implemented. Optional controllers are executed after the two before-mentioned controllers.
It would be particularly interesting to enable distributed control of wind turbines in a wind
farm set-up [10]. The Simulator for Offshore Wind Farm Applications (SOWFA) already has a
build in super-controller that is able to communicate with local wind turbine controllers [11].
In later stages, this coupling between the DRC and SOWFA will be made possible.
The DRC source code is publicly available under terms of the GNU Public License version 3
as a repository on the Data-Driven Control GitHub website

https://github.com/TUDelft-DataDrivenControl/DRC_Fortran

An advantage of using a public repository is the ability to share the controller by only referring
to the Git commit hash (controller version) and parameter file. This improves reproducibility
of the performed research, as all changes between commits remain accessible.
Along with the source code, a makefile and a Visual Studio project are supplied for
compilation with gfortran [12] or Intel Visual Fortran [13] under Linux and/or Windows,

3
The Science of Making Torque from Wind (TORQUE 2018) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1037 (2018)
1234567890 ‘’“” 032009 doi:10.1088/1742-6596/1037/3/032009

for creation of .so and .dll library files. Baseline parameter configuration files are supplied for
the NREL5MW and DTU 10-MW reference wind turbines.

2.2. Filter module

The DRC comes with a collection of frequently used filters, which are contained in the filter
module Filters.f90. In this section, the continuous time representations of the included filters
are discussed. The filters included are discretized using the bilinear transformation, also known
as Tustin’s method, which is defined by

2 1 − z −1
s= , (1)
DT 1 + z −1

where DT is the sampling time and z −1 the discrete-time unit delay operator. This transformation
maps every point of the frequency response of the continuous-time filter to a corresponding point
in the frequency response of the discrete-time filter [14]. The filters are not bound to a predefined
sampling time, as this variable is taken as an input from the simulation. The general function-call
syntax is given by filter(InputSignal, DT, Opt1, Opt2, ..., iStatus, reset, inst).
To avoid transients, the inputs iStatus and a reset are used to reinitialize the filter states
(initial condition) to the current input signal. All functions return a single real filtered output
signal, with a default steady-state gain of 0 dB.
Fortran is not an object-oriented language, which would be especially convenient when
working with multiple instances of the same subroutine. For example, one might want to use
a certain filter for different purposes without redefinition. To solve this problem, the DRC
incorporates an implementation to create multiple instances of functions and subroutines and
stores the instance ID’s in the ObjectInst type. The instance-ID of the filter is determined by
the last input argument inst, which is automatically incremented in a filter function call.
First/second-order low-pass filter: first- and second-order low-pass filters pass signals
with frequencies lower than the cross-over frequency, but attenuate signal components above
this frequency. This filters are represented by

ωc ωc2
L1 (s) = , L2 (s) = (2)
s + ωc s2 + 2ζωc s + ωc2

where ωc = 1/τc represents the corner frequency, and ζ is the damping coefficient.
First-order high-pass filter: a first-order high-pass filter passes signals with frequencies
higher than the cut-in frequency, but attenuates signal components below this frequency. This
filter is represented by
s
H1 (s) = . (3)
s + ωc

Notch filter: a notch filter, often also referred to as a band-stop or a band-rejection filter,
passes most of the frequencies but attenuates in a very specific predefined interval. The filter is
represented by

s2 + 2β1 ωn s + ωn2
N = , (4)
s2 + 2β2 ωn s + ωn2

where the magnitude and ratio between β1 and β2 determines the amplification/attenuation
magnitude and width, respectively.

4
The Science of Making Torque from Wind (TORQUE 2018) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1037 (2018)
1234567890 ‘’“” 032009 doi:10.1088/1742-6596/1037/3/032009

Inverted-notch filter, with decreasing slopes: an inverted notch with decreasing slopes
on both sides of ωn provides extra attenuation of frequencies outside the filtering frequency, for
both higher and lower frequencies. The transfer function describing this filter is given as
s
Ns (s) = , (5)
(Q/ωn )s2 + s + Qωn

where Q is the quality factor coefficient: lower and higher values of Q give a broader and sharper
selection of the frequency ωn , respectively.

2.3. Function module

The DRC controller is supplied with frequently used functions which are contained in the
function module Functions.f90. The included functions are described briefly in this section.
Value/signal saturation: the saturate() function saturates a given input signal to a
upper and lower value, and returns a real output signal.
Signal rate limiter: a signal rate with respect to time can be bounded by the ratelimit()
function. The minimum and maximum rate bounds are defined on a per-second basis.
Proportional-Integral (PI) controller: an object-based PI-controller is included by the
function PIController(). Saturations are included on both the integral action as the output
signal to provide anti-integrator wind-up and signal saturation capabilities.
1D-interpolation: a one-dimensional interpolation function interp1d() is included to
perform evaluation of a value in the interval of a one-dimensional table, of which the x-data
should be monotonically increasing.

3. Baseline control overview

This section describes the baseline control implementations and strategies included in the DRC.
First, a state-machine determining the state of the operational wind turbine is described in
Section 3.1. Then, Section 3.2 describes the baseline pitch and torque controllers. Next,
individual pitch control (IPC) and yaw-rate control are outlined in Sections 3.3 and 3.4.

3.1. State-machine
The purpose of the global state-machine is to determine the operational state of the wind
turbine, and is included as a subroutine in the function module. Inside this function, separate
state-machines are included for pitch and torque control. The operational state is determined
by comparing measured turbine quantities and control signal values to controller settings
defined in the ControllerParameters.in configuration file. Figure 2 gives an overview
of the different wind turbine operating regions, indicating internal controller variables and
configuration parameters.

3.2. Baseline pitch and torque control

This section presents the pitch and torque control implementations included in the DRC. The
controllers use a common generator speed measurement GenSpeedF filtered by a second-order
low-pass filter with cut-off frequency CornerFreq, to calculate the error from the reference
parameter. Both controllers act on individual generator speed set points defined by VS RefSpd
and PC RefSpd for torque and pitch control.
At the end and beginning of regions 1 and 2.5, two PI torque controllers are implemented
to regulate the rotor speed towards the optimal below-rated torque path, and to above-rated
operating conditions. The controller gains are defined by VS KP and VS KI. The controllers
continuously calculate the below- and above-rated torque references GenBrTq and GenArTq,
and act on different error signals VS SpdErrBr and VS SpdErrAr. As shown in Figure 2, the

5
The Science of Making Torque from Wind (TORQUE 2018) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1037 (2018)
1234567890 ‘’“” 032009 doi:10.1088/1742-6596/1037/3/032009

Figure 2. Torque control strategies implemented in the DRC. All variables regarding torque
control are indicated by their respective names present in the control parameter file. For above-
rated operation (Region 3), the control strategy can be configured to be either in constant torque
of constant power mode by the parameter VS ControlMode.

maximum torque controller saturation VS ArSatTq can be set independently to provide in less
frequent switches between torque and pitch control. The torque signal change rate is limited
by VS MaxRat, and the physical minimum and maximum torque constraints are defined by
VS MinTq and VS MaxTq. The cut-in speed for below-rated torque control (Region 2) is defined
by VS MinOMSpd, and the optimal mode-gain for tracking Cp,max is set by VS Rgn2K.
For power regulation during above-rated operation the torque controller can be configured to
either deliver a constant torque signal, or actively change the torque signal to obtain a constant
power output by setting the parameter VS ControlMode to 0 or 1, respectively. When constant
torque control is selected, the generator torque is VS RtTq; for constant power tracking, the
torque signal varies subject to the instantaneous generator speed and the electrical power set
point defined by VS RtPwr. The generator efficiency VS GenEff must be defined and should
match the value specified in the turbine drivetrain model parameters.
The baseline pitch controller is implemented as a gain-scheduled PI-controller. The
proportional and integral gains PC GS KP and PC GS KI, are scheduled on the commanded
pitch angle of the previous controller iteration. The gain information is defined in the
ControllerParameters.in parameter file as function of pitch angles PC GS angles, and is
interpolated in each time step by the interp1d() function.
Pitch control is active in all operating regions. However, based on the turbine operational
state PC State determined by the state machine, the upper saturation bound is limited to the
fine-pitch angle PC FinePit or the maximum physical pitch angle PC MaxPit. The pitch control
signal rate is limited by a lower and upper bound PC MinRat and PC MaxRat. The PC Switch
parameter is added to the fine-pitch angle PC FinePit parameter, and the summed value is used
to indicate when the torque controller should switch to region 3.

6
The Science of Making Torque from Wind (TORQUE 2018) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1037 (2018)
1234567890 ‘’“” 032009 doi:10.1088/1742-6596/1037/3/032009

IPC_ControlMode = 1
Y_ControlMode = 0 or 1

rootMOOP(1) PitComIPC(1)
axTOut axTIn
PI-controller Coleman
rootMOOP(2) Coleman PitComIPC(2)
axYOut
IPC_KI axYIn
Transform
Transform() Inverse()
IPC_IntSat

rootMOOP(3) PitComIPC(3)

IPC_aziOffset

Figure 3. Individual pitch control for blade fatigue load reductions. The out-of-plane blade
root moments are transformed in a tilt- and yaw-axis by a 1P Coleman transformation. After PI-
control, the resulting pitch angles are transformed back by an inverse Coleman transformation
to obtain IPC pitch signals to mitigate 1P fatigue loadings.

3.3. Individual pitch control

An individual pitch control (IPC) implementation based on the Coleman transformation [16] is
included in the DRC for two distinct purposes: for fatigue load reductions and blade moment
yaw-control (yaw-by-IPC). The two IPC modes cannot be activated simultaneously. Both
implementations are described in this section.
IPC for fatigue load reductions: a schematic overview of IPC for blade moment fatigue
load reductions is presented in Figure 3. Setting IPC ControlMode to 1 enables IPC to reduce
fatigue out-of-plane blade root moments, by adding contributions to the pitch control signals.
The implementation focuses on attenuation of the 1P blade load harmonic. The measured
blade root out-of-plane moments, together with the rotor azimuth angle, are taken as input
to the forward Coleman transformation, resulting in non-rotating rotor tilt- and yaw-moments.
Subsequently, the two signals are integrated with a gain IPC KI, and the integrated signal is
saturated to IPC IntSat to prevent excessive pitch contributions. Finally, the two pitch signals
in the non-rotating frame are subject to a reverse Coleman transformation to obtain actual

IPC_ControlMode = 0
Y_ControlMode = 2 axTIn = 0 PitComIPC(1)

Coleman PitComIPC(2)
Transform
Inverse()
PI-controller
Y_MErrSet + axYIn
Y_IPC_KP PitComIPC(3)
Y_IPC_KI
Y_IPC_IntSat
Y_M
IPC_aziOffset

Figure 4. A IPC yaw control implementation for a wind turbine where the nacelle is mounted
on the tower, in a free-damped fashion. The transformed tilt pitch angle is nullified, and the yaw
angle is actively controlled by the error between the set point and measured yaw misalignment.

7
The Science of Making Torque from Wind (TORQUE 2018) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1037 (2018)
1234567890 ‘’“” 032009 doi:10.1088/1742-6596/1037/3/032009

Y_ErrLPFFast Y_AccErr
Y_omegaLPFast ∫sign(u)u2 dt

Y_MErrSet + Y_MErr Y_ErrLPFSlow No

Y_omegaLPSlow |Y_AccErr|>
Y_ErrThresh

Y_M
Reset filters
and integrator Yaw to Yes
Y_ErrLPFSlow

Figure 5. Yaw rate control uses the error between the misalignment set point and the measured
misalignment with the dominating wind direction to intermittently perform yaw manoeuvres.

implementable pitch signals. A phase offset IPC aziOffset can be added to the azimuth angle
in the reverse transformation.
Yaw-by-IPC: IPC can be configured to act in a yaw-by-IPC set-up, by setting
Y ControlMode to 2. By doing so, the blades induce a yaw moment on the entire rotor to
actively regulate or track a yaw misalignment set point. With yaw-by-IPC, a 1P contribution is
added to the pitch signals resulting in a yaw moment over the entire rotor. Normally, this type
of control is present in downwind wind turbines, where the nacelle is mounted in a free-damped
fashion on the tower support structure [9, 15]. A schematic overview of the implementation is
presented in Figure 4. In this figure it is shown that the yaw misalignment error Y MErr, which
is the difference between measured yaw misalignment Y M and set point Y MErrSet, is fed to a
PI-controller with proportional and integral gains Y IPC KP and Y IPC KI. The control output
signal is saturated by Y IPC IntSat.

3.4. Yaw-rate control

The yaw-rate controller uses measurements from a wind vane located downwind, i.e., seen
from upwind the vane is positioned behind the rotor and tower. The wind vane measures
the nacelle yaw-misalignment with respect to the dominating wind direction, but does not give
information on the absolute nacelle orientation. Yaw motors with a fixed yaw-rate are used for
yaw movements. The yaw-rate control implementation does not provide continuous alignment,
but intermittently aligns the turbine nacelle when a predefined threshold is exceeded.
The implementation adapted from [8] and schematically depicted in Figure 5, is slightly
adjusted to allow for yaw-angle offsets. Two distinct low-pass filters on the yaw-error signal
Y MErr are employed. A fast low-pass filter with a higher corner-frequency Y omegaLPFast is used
to determine when the turbine should yaw according to some predefined threshold Y ErrThresh.
The squared value of the yaw-error is used to obtain a higher penalization for larger yaw
excursions, and is integrated into Y AccErr. Once the integrated error exceeds the predefined
threshold, the value from the low-pass filter with a lower corner-frequency Y omegaLPSlow is
taken to determine the yaw time using the fixed angular yaw speed parameter Y Rate.

4. Results
This section presents the performance and behavior of the DRC baseline controller in
combination with OpenFAST v1.0.0-x64. The DTU10MW wind turbine model introduced in
Section 1, is used for evaluation purposes in two distinct simulations.

8
The Science of Making Torque from Wind (TORQUE 2018) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1037 (2018)
1234567890 ‘’“” 032009 doi:10.1088/1742-6596/1037/3/032009

Constant power
Constant torque

Figure 6. Simulation of the DTU10MW with constant power and constant torque control,
subject to a wind profile with mean total velocity of 12 m/s and IEC turbulence characteristic
and type NTM-1A.

First, the wind turbine is subject to a IEC-1A Normal Turbulence Model (NTM) wind profile
with a mean wind speed of 12 m/s. Using this wind profile, the torque controller of the DRC is
configured in constant power and constant torque mode, and the results are presented in Figure 6.
It is shown that for both cases the controller handles below- and above-rated operation well by
switching between torque and pitch control. Power regulation is improved using the constant
power control strategy, at cost of continuous adjustment of the generator torque. The effect on
pitch control and the generator speed for the two strategies is negligible.

Yaw-by-IPC
Yaw rate control

Figure 7. Simulation of the yaw-rate and yaw-by-IPC control implementation, tracking

a constant yaw-misalignment set point of 20 deg. The yaw-rate controller is only active
intermittently, while the yaw-by-IPC controller continuously regulates the free-yaw turbine
around the misalignment set point using distinct blade pitch angles. For the latter case, the
blade pitch rate does not exceed 3 deg/s.

9
The Science of Making Torque from Wind (TORQUE 2018) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1037 (2018)
1234567890 ‘’“” 032009 doi:10.1088/1742-6596/1037/3/032009

Secondly, a simulation is performed to showcase the performance of the yaw-rate and yaw-by-
IPC controllers by tracking a misalignment set point of 20 deg perpendicular to the rotor plane.
Results are shown in Figure 7, where the same wind profile as defined in the first simulation
is used. For yaw-by-IPC control, the yaw spring-stiffness of the DTU-10MW model is removed
while retaining the yaw-damping between nacelle and tower.

Conclusion and outlook

The DRC provides an open and community-driven wind turbine baseline controller, and aims
in being the reference controller for evaluation of new control algorithms. The controller is
set up in such a way that it is applicable to each wind turbine model in frequently used wind
turbine simulation software. Only a single parameter file is needed to configure the controller,
which abandons the need for recompilation under a change in controller settings. Because of
the modular set-up, the existing baseline control implementations are easily replaced, which
enables for convenient comparison, evaluation and reproducibility of new algorithms. Although
the baseline controller is build on basic PI controllers, more advanced control methods can be
implemented by incorporating external Fortran libraries. The next step is to incorporate a
general applicable wind speed estimator, enabling closed-loop tip-speed ratio tracking during
below-rated operation. In later stages, the DRC functionality will be further extended enabling
a distributed control set-up in a wind farm setting.

Acknowledgments
We thank J. Govers and J. Hoorneman (TU Delft) for their dedication initiating the DRC
project. We acknowledge CL-Windcon for their contributions improving the code.

References
[1] Jonkman J, Butterfield S, Musial W and Scott G 2009 Definition of a 5-MW reference wind turbine for
offshore system development Tech rep. NREL, Golden, CO.
[2] Bak C, Zahle F, Bitsche R, Kim T, Yde A, Henriksen L, Andersen P, Natarajan A and Hansen M 2013 Design
and performance of a 10MW wind turbine H, J. Wind Energy
[3] DNV-GL 2018 Bladed; https://www.dnvgl.com/energy
[4] NWTC 2018 OpenFAST http://openfast.readthedocs.io
[5] Hansen MH and Henriksen LC 2013 Basic DTU wind energy controller DTU Wind Energy. DTU Wind Energy
E, no. 0028
[6] Garrad Hassan & Partners Ltd Bladed User Manual version 4.2
[7] Lahey T M and Ellis T 1994 Fortran 90 programming (Addison-Wesley Longman Publishing Co., Inc.)
[8] Kragh K and Fleming P 2012 Rotor Speed Dependent Yaw Control of Wind Turbines Based on Empirical
Data AIAA 2012 1 - 9
[9] Van Solingen E, Beerens J, Mulders S, De Breuker R and Van Wingerden JW 2016 Control design for a
two-bladed downwind teeterless damped free-yaw wind turbine Mechatronics 36 7796
[10] Boersma S, Doekemeijer BM, Gebraad PMO, Fleming PA, Annoni J, Scholbrock AK, Frederik JA and Van
Wingerden JW 2017 A tutorial on control-oriented modeling and control of wind farms IEEE ACC 2017
1 - 18
[11] Fleming P, et al. 2013 SOWFA Super-Controller: A High-Fidelity Tool for Evaluating Wind Plant Control
Approaches (No. NREL/CP-5000-57175). NREL, Golden, CO
[12] Project G 2018 GNU Fortran (GFortran) https://gcc.gnu.org/fortran/
[13] Intel 2018 Intel Fortran Compiler https://software.intel.com/en-us/fortran-compilers
[14] Oppenheim A V 1999 Discrete-time signal processing (Pearson Education India)
[15] Schorbach V and Dalhof P 2012 Two bladed wind turbines: antiquated or supposed to be resurrected
Proceedings of the EWEA Conference
[16] Bir G 2008 Multi-blade coordinate transformation and its application to wind turbine analysis AIAA 46th

10